From f466739b09be1baffd13110b431ecd8057a40771 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Mon, 24 Aug 2020 13:43:13 -0700
Subject: WIP: expanding query parsing

---
 src/lib.rs | 172 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 165 insertions(+), 7 deletions(-)

diff --git a/src/lib.rs b/src/lib.rs
index 1ba6207..615e4ed 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,35 +1,193 @@
 
 use serde::{Serialize, Deserialize};
+use std::collections::HashMap;
 
 #[derive(Serialize, Deserialize, Debug)]
 pub struct ApiRequest {
     pub method: String,
     pub path_and_query: String,
-    pub body: Option<ApiBody>,
+    pub body: Option<SearchBody>,
 }
 
+/*
+
+URL query parameters:
+
+    pub allow_no_indices: Option<bool>,
+    pub allow_partial_search_results: Option<bool>,
+    pub batched_reduce_size: Option<u32>,
+    pub ccs_minimize_roundtrips: Option<bool>,
+    pub docvalue_fields: Option<String>, // array of strings, comma-separated
+    pub expand_wildcards: Option<String>,
+    pub explain: Option<bool>,
+    pub from: Option<u32>,
+    pub ignore_throttled: Option<bool>,
+    pub ignore_unavailable: Option<bool>,
+    pub max_concurrent_shard_requests: Option<u32>,
+    pub pre_filter_shard_size: Option<u32>,
+    pub preference: Option<String>,
+    pub q: Option<String>
+    pub request_cache: Option<bool>,
+    pub rest_total_hits_as_int: Option<bool>,
+    pub routing: Option<String>,
+    pub scroll: Option<String>, // string is "time value"
+    pub search_type: Option<String>,
+    pub seq_no_primary_term: Option<bool>,
+    pub size: Option<u32>,
+    pub sort: Option<String>, // array of strings, comma-separated
+    pub _source: Option<bool>, // TODO: bool or string
+    pub _source_excludes: Option<String>, // array of strings, comma-separated
+    pub _source_includes: Option<String>, // array of strings, comma-separated
+    pub stats: Option<String>,
+    pub stored_fields: Option<String>, // array of strings, comma-separated
+    pub suggest_field: Option<String>,
+    pub suggest_text: Option<String>,
+    pub terminate_after: Option<u32>,
+    pub timeout: Option<String>, // string is "time units"
+    pub track_scores: Option<bool>,
+    pub track_total_hits: Option<bool>, // XXX: bool or integer
+    pub typed_keys: Option<bool>,
+    pub version: Option<bool>,
+*/
+
+// https://www.elastic.co/guide/en/elasticsearch/reference/current/search-search.html
 #[derive(Serialize, Deserialize, Debug)]
-pub struct ApiBody {
+pub struct SearchBody {
     pub query: Option<ApiQuery>,
+    pub highlight: Option<ApiHighlight>,
+    pub collapse: Option<ApiCollapse>,
+    pub post_filter: Option<ApiQuery>, // TODO: leaf query only?
+    pub rescore: Option<ApiRescore>, // TODO: single or an array of rescore objects
+    // script_fields disabled
+
+    // https://www.elastic.co/guide/en/elasticsearch/reference/current/sort-search-results.html
+    // pub sort: Option<Vec<SortThing>>, // XXX: array of object or string
+
+    pub slice: Option<ApiSlice>,
+    pub stored_fields: Option<String>, // array of strings, or "_none_"
+
+    // overlap with URL query parameters
+    pub docvalue_fields: Option<Vec<String>>, // XXX: array of strings or objects? {field, format} in the objects
+    pub explain: Option<bool>,
     pub from: Option<u32>,
+    pub min_score: Option<f64>,
+    pub seq_no_primary_term: Option<bool>,
     pub size: Option<u32>,
-    pub sort: Option<String>, // XXX
-    pub slice: Option<String>, // XXX
+    pub _source: Option<bool>, // XXX: bool, string, or object
+    pub terminate_after: Option<u32>,
+    pub timeout: Option<String>, // string is "time units"
+
+    // not enumerated in direct docs, but seems to be allowed?
+}
+
+#[derive(Serialize, Deserialize, Debug)]
+pub struct ApiSlice {
+    id: u32,
+    max: u32,
+    field: Option<String>,
+}
+
+#[derive(Serialize, Deserialize, Debug)]
+pub struct ApiRescore{
+    pub query: Option<ApiQuery>,
+    pub window_size: Option<u32>,
 }
 
 #[derive(Serialize, Deserialize, Debug)]
 pub enum ApiQuery {
+    // compound queries
+    #[serde(rename = "bool")]
+    Bool(HashMap<String, QueryFieldOrString>),
+    #[serde(rename = "boosting")]
+    Boosting {positive: Box<ApiQuery>, negative: Box<ApiQuery>, negative_boost: f64},
+    #[serde(rename = "constant_score")]
+    ConstantScore {filter: Box<ApiQuery>, boost: Option<f64>},
+
+    // fulltext (leaf) queries
+
+    // term-level (leaf) queries
     #[serde(rename = "match")]
-    Match(MatchQuery),
+    Match(HashMap<String, QueryFieldOrString>),
+    #[serde(rename = "match_phrase")]
+    MatchPhrase(HashMap<String, QueryFieldOrString>),
+    #[serde(rename = "query_string")]
+    QueryString(QueryField),
+
+    // other
+    #[serde(rename = "nested")]
+    Nested(NestedQuery),
+    #[serde(rename = "rescore_query")]
+    Rescore(Box<ApiQuery>),
+}
+
+#[derive(Serialize, Deserialize, Debug)]
+pub struct ApiHighlight{
+    // TODO: fields could also be an array of strings?
+    fields: HashMap<String, HighlightField>,
+
+    #[serde(flatten)]
+    settings: HighlightField,
 }
 
 #[derive(Serialize, Deserialize, Debug)]
-pub struct MatchQuery {
-    message: QueryField,
+#[serde(untagged)]
+pub enum QueryFieldOrString {
+    Object(QueryField),
+    String(String),
 }
 
 #[derive(Serialize, Deserialize, Debug)]
 pub struct QueryField{
     query: String,
     fuzziness: Option<String>,
+    slop: Option<u32>,
+    boost: Option<f64>,
+}
+
+#[derive(Serialize, Deserialize, Debug)]
+pub struct NestedQuery{
+    path: String,
+    query: Box<ApiQuery>,
+    score_mode: Option<String>,
+    ignore_unmapped: Option<bool>,
+}
+
+// https://www.elastic.co/guide/en/elasticsearch/reference/current/highlighting.html
+#[derive(Serialize, Deserialize, Debug)]
+pub struct HighlightField{
+    boundary_chars: Option<String>,
+    boundary_max_scan: Option<u32>,
+    boundary_scanner: Option<String>,
+    boundary_scanner_locale: Option<String>,
+    encoder: Option<String>,
+    force_source: Option<bool>,
+    fragmenter: Option<String>,
+    fragment_offset: Option<u32>,
+    fragment_size: Option<u32>,
+    highlight_query: Option<ApiQuery>,
+    matched_fields: Option<Vec<String>>,
+    no_match_size: Option<u32>,
+    number_of_fragments: Option<u32>,
+    order: Option<String>,
+    phrase_limit: Option<u32>,
+    pre_tags: Option<Vec<String>>,
+    post_tags: Option<Vec<String>>,
+    require_field_match: Option<bool>,
+    tags_schema: Option<String>,
+    #[serde(rename = "type")]
+    highlight_type: Option<String>,
+}
+
+#[derive(Serialize, Deserialize, Debug)]
+pub struct ApiCollapse{
+    field: String,
+    inner_hits: Option<InnerHits>,
+}
+
+#[derive(Serialize, Deserialize, Debug)]
+pub struct InnerHits {
+    from: Option<u32>,
+    size: Option<u32>,
+    sort: Option<Vec<String>>,
+    name: Option<String>,
 }
-- 
cgit v1.2.3